Modelpublic

Abiray/harrier-oss-v1-27b-GGUF

sourceHugging Faceupdated 6mo agoView on Hugging Face

5likes891downloads

Model Card

Harrier OSS V1 27B - GGUF

This repository contains GGUF quantized formats of the massive 27-billion parameter `microsoft/harrier-oss-v1-27b` embedding model.

By utilizing these GGUF quants, you can run state-of-the-art semantic representation and text embedding generation on standard consumer hardware and CPUs.

Available Quantizations

File Name	Bit Depth	Description
harrier-27b-Q8_0.gguf	8-bit	Highest quality, virtually indistinguishable from FP16. Recommended if you have 32GB+ RAM.
harrier-27b-Q6_K.gguf	6-bit	Excellent balance of quality and size.
harrier-27b-Q5_K_M.gguf	5-bit	Great middle ground for memory constraints.
harrier-27b-Q4_K_M.gguf	4-bit	Smallest footprint (~16.6 GB). Runs comfortably on machines with 24GB or 32GB of RAM.